I would love it if I could get a baseline on what people are experiencing with client crash rates. Fortunatly in ND6 you can monitor the frequency of crashing with Automatic Diagnostic Collection. I have had it turned on in a shop with about 2000 people upgraded and this is what I have found:
20-40 NSD’s generated per day - this is 1-3% of users per day
I have done a pretty thorough analysis of these NSD’s and this is what I’ve found
Out of 590 NSD’s
173 were caused by a known issue in Local Scheduled Agents being enabled
30 were caused by an older version of Blackberry Manager installed
30 Were caused by another known issue related to authentication but that isn’t reproducable and doesn’t have a fix
2 sets of between 30 and 50 are the same NSD pattern which hasn’t been identified in KBase
The rest are a random smattering of groups of between 1 and 10 NSD with matching patterns.
Anyone else who is looking at this and has a feel for frequency please post here.
Subject: SURVEY How often is the client crashing - My experience and analysis
I’m really impressed with your analysis of NSD logs. Do you analyze each new NSD manually? Do you keep some kind of database with the information, so that when a new one comes in you can match it as another instance of problem X?
Subject: RE: SURVEY How often is the client crashing - My experience and analysis
Here is how I do it, it is a bit of a pain but I’ve been getting better at it. I keep a local replica of the diagnostic db which I full text index.
I pick an NSD to open and I search for “Fatal” and grab some text from the top of the stack like namgrdll._IsDbCurrentlySecured@0+5 (from the amgr crash)
Then I do a full text search back in the db on that value and start looking at the results. Sometimes they don’t match and just are calling the same thing but lots of the time I do find matches.
When I find a match I take all of the matching documents and drag them into a folder to take them out of the population that I’m working with to make future searches easier.
Subject: SURVEY How often is the client crashing - My experience and analysis
Our standard desktop uses the JavaScript setTimeOut to refresh the content of a frame every x seconds, so as to deliver up to date rich text (company news etc) to the desktop. Basically the document in the frame calls .setlocation on its own window every x seconds. We have found that the resources used by the document are not released and this results in a redbox after n iterations, or a couple of hours of use. We don’t have a solution yet but I’m sure we’ll arrive at one soon.
I installed R6 from the download and have found it very stable (but I don’t use the standard desktop and didn’t install it on top of R5).